73MIS Assessment Methods and Metrics
Da Vinci Application Programming Interface
The da Vinci robot stores vast amount of user information, but in a “black box” of
proprietary algorithms and inaccessible data, both raw and processed. There is cur-
rently no way, as with the simulators, to download operative data to an external
drive.
The da Vinci’s application programming interface (API) is a bridge between this
data and the outside world. Live information can be streamed through an Ethernet
connection to another computer containing kinematic data of the user. This data
includes how the patient cart joint angles were set up, how the master controllers are
positioned and moved, Cartesian positions of the master controllers, patient side
instruments and endoscope, and velocity of the master controllers and patient side
joints. Any actions taken by the surgeon at the console are also saved, including data
from the head sensor trigger, and standby and ready button usage, as well as master
clutch, camera control, and arm swap pedals.
API data is not openly available. Those who are interested in using it must first
enter a legal agreement with ISI. The agreement defines liability and intellectual
property rights and is only awarded if specific conditions are met. Intuitive requires
that research is designed for long-term results that is cohesive with current internal
research, that the researchers have experience with the tasks at hand and work in a
supportive clinical environment, and that the researchers and clinic can communi-
cate effectively [ 23 ]. Only then will the API interface be activated, and onsite train-
ing by intuitive on how to best utilize the resource can proceed.
API motion data is transferred over a frequency ranging from 10 to 100 Hz at 334
different data measurements. This massive amount of information can be utilized in
a number of ways for skills analysis [ 23 ].
Kumar et al. employed the API to quantify expert proficiency and differentiation
from nonexpert task completion by measuring master controller movement while
instruments were clutched [ 24 ]. In doing so, they measured only operative skills.
Operative skills are those involved with how the operator (surgeon) interacts with
the machine (robot). Typical Halsteadian training focuses instead on procedural
skills (like suturing) or the surgical technique. By eliminating measurement of other
skills, they were able to directly assess how a person was utilizing the technology,
rather than their overall adeptness at surgery.
The data measured from master movement during clutching were translated into
a vector with a Cartesian position plot. Analysis of the vectors led to 87–100% accu-
racy in identifying expert and nonexpert trainees [ 24 ]. Setting thresholds on the vec-
tor values noted at expert values for trainees would ensure an objective and
quantifiable means of assessment for robotic operational skills. The assessment
could even take place without interfering with other training or clinical schedules, as
the API is integrated into the da Vinci system with minimal additional hardware.
Da Vinci’s API can do more than assess proficiency. Several teams have used the
technology to analyze what movements define a particular task. Lin et al. asked
5 Performance Assessment in Minimally Invasive Surgery